7
Proposed Aerial Triangulation Techniques*t MICHAEL G. MISULIA, Technical Developments Staff, Army Map Service, Washington, D. C. ABSTRACT: Several novel aerial triangulation techniques utilizing stereo- photogrammetric instrumentation are advanced for establishing control in areas containing insufficient geodetic control for producing topographic maps. The approaches advanced are designed to yield increased accuracies and to gain improved reliability of results. Because of the complexity and limited range of airborne devices for recording positional information and the time-consuming operations required to reduce and correlate these data with aerial triangulation processes, special attention is directed to the need for an improved system for securing and handling these data. A n approach to this problem, which may alltviate some of these limitations, is presented. INTRODUCTION T HERE is a need for developing improved photogrammetric techniques for map- ping areas where control is sparse or non- existent, in order to offset the high costs and excessive amount of time normally re- quired to establish control by conventional ground-survey operations. The recent in- troduction of electronic survey instrumenta- tion such as the Telurometer, although capable of substantially reducing these cost and ti me factors, should not be considered as a final cure-all. This is so because of the need for a dense network of control in stereoplotting operations, and the inaccessi- bility of ground teams to certain regions due to terrain or military obstacles. During the past few years, many different methods of securing supplemental control for photogrammetric mapping have been suggested. Basically, two avenues of ap- proach are possible; first, securing control data coincident with the photographic mis- sion; and second, securing control data prior to or after receipt of the aerial photography. From a logistic and time standpoint, the coincident approach is usually desirable since the required control data are procured during the photo mission and initiation of the mapping project may be expedited. Examples of this are the APR and HIRA - SHORAN Systems which are employed in conjunction with aerial photography to se- cure positional data. Insufficient data have yet been secured to firmly establish the ac- curacy of these systems in various latitudes and terrain and under different operational conditions. Use of these systems necessitates the need for rather elaborate electronic equipment and subsequent reduction of data for correlation with aerial triangulation processes. Knowledge of meteorological conditions is also essential if proper adjust- ments can be made to the data obtained; otherwise serious errors in the results achieved may be introduced. However, even when these corrections are applied, appreci- able errors may still be present because of the random nature of atmospheric condi- tions and terrain characteristics, over large areas in various portions of the globe. Present photogrammetric aerial triangu- lation techniques are not wholly satisfactory for establishing control in areas which must be bridged for long distances without inter- mediate control or where the control must be extended into an area without recourse to any tie-in points (see Figure 1). Results obtained in cantilever tests, utilizing various types of precision stereoplotting instru- ments, clearly indicate that the resultant accuracy is very indefinite. Even under * The material and opinions expressed are those "proposed" by the author on the basis of investi- gations to date and should not be construed as conclusions of the Chief of Engineers. t Presented at the Society's 24th Annual Meeting, Hotel Shoreham, Washington, D. c., March 28, 1958. 72S

Proposed Aerial Triangulation Techniques

  • Upload
    others

  • View
    2

  • Download
    0

Embed Size (px)

Citation preview

Page 1: Proposed Aerial Triangulation Techniques

Proposed Aerial TriangulationTechniques*t

MICHAEL G. MISULIA,

Technical Developments Staff, Army Map Service,Washington, D. C.

ABSTRACT: Several novel aerial triangulation techniques utilizing stereo­photogrammetric instrumentation are advanced for establishing control inareas containing insufficient geodetic control for producing topographicmaps. The approaches advanced are designed to yield increased accuraciesand to gain improved reliability of results. Because of the complexity andlimited range of airborne devices for recording positional information andthe time-consuming operations required to reduce and correlate these datawith aerial triangulation processes, special attention is directed to the needfor an improved system for securing and handling these data. A n approachto this problem, which may alltviate some of these limitations, is presented.

INTRODUCTION

T HERE is a need for developing improvedphotogrammetric techniques for map­

ping areas where control is sparse or non­existent, in order to offset the high costsand excessive amount of time normally re­quired to establish control by conventionalground-survey operations. The recent in­troduction of electronic survey instrumenta­tion such as the Telurometer, althoughcapable of substantially reducing these costand ti me factors, should not be consideredas a final cure-all. This is so because of theneed for a dense network of control instereoplotting operations, and the inaccessi­bility of ground teams to certain regions dueto terrain or military obstacles.

During the past few years, many differentmethods of securing supplemental controlfor photogrammetric mapping have beensuggested. Basically, two avenues of ap­proach are possible; first, securing controldata coincident with the photographic mis­sion; and second, securing control data priorto or after receipt of the aerial photography.From a logistic and time standpoint, thecoincident approach is usually desirablesince the required control data are procuredduring the photo mission and initiation ofthe mapping project may be expedited.Examples of this are the APR and HIRA -

SHORAN Systems which are employed inconjunction with aerial photography to se­cure positional data. Insufficient data haveyet been secured to firmly establish the ac­curacy of these systems in various latitudesand terrain and under different operationalconditions. Use of these systems necessitatesthe need for rather elaborate electronicequipment and subsequent reduction of datafor correlation with aerial triangulationprocesses. Knowledge of meteorologicalconditions is also essential if proper adjust­ments can be made to the data obtained;otherwise serious errors in the resultsachieved may be introduced. However, evenwhen these corrections are applied, appreci­able errors may still be presen t because ofthe random nature of atmospheric condi­tions and terrain characteristics, over largeareas in various portions of the globe.

Present photogrammetric aerial triangu­lation techniques are not wholly satisfactoryfor establishing control in areas which mustbe bridged for long distances without inter­mediate control or where the control mustbe extended into an area without recourseto any tie-in points (see Figure 1). Resultsobtained in cantilever tests, utilizing varioustypes of precision stereoplotting instru­ments, clearly indicate that the resultantaccuracy is very indefinite. Even under

* The material and opinions expressed are those "proposed" by the author on the basis of investi­gations to date and should not be construed as conclusions of the Chief of Engineers.

t Presented at the Society's 24th Annual Meeting, Hotel Shoreham, Washington, D. c., March28, 1958.

72S

Page 2: Proposed Aerial Triangulation Techniques

proaches are only in the embryo stage, andfurther study and evaluation are neededbefore they can be considered feasible formapping applications. Lt. Col. K. Evinayof the Turkish Army, during his stay at theArmy Map Service, graciously offeredmany helpful suggestions in formulatingthese techniques.

FIG. 3. Cantilever extension-positional errors.

! "Signalizing ~nvestigation" by EngineersHlawaty and StIckIer, Photogrammetria XII1955-56, #4, pp. 236. ' ,

SLAVE AIRCRAFT

It is proposed that a slave aircraft beflown simultaneously with two masteraerial photographic planes, at a relativelvmuch lower altitude and located approxi­mately between the photo planes. The mas­ter aircrafts may either be in line of flight(Figure 4a) or in parallel flights (Figure 4b).

From economic and elapsed time con­siderations, the parallel-flight case would bemore desirable because two flights of photog­raphy are obtained simultaneously. Theslave aircraft would be rigged in a mannersimilar to planes towing two advertisingsi.gns (see Figure 5a). For simplicity, thedIstances between the towed targets wouldbe equidistant. The distance from the tar­ge~s would be dependent upon the flightaltItude of the photo-mapping planes. Toobtain optimum results, it would be desir­able to have the total target length ap­proximately equal to the stereo modelcoverage. For practical considerations, how­ever, much smaller distances could be used.The shape, size, and color of the targetswould also be designed to achieve optimumstability in flight and plotting-instrumentsighting.

Ground control targets were tested! withWild RC-5 Aviotar and Zeiss-Kramer MRKTopar mapping cameras at 1 :5,000 and1: 10,000 scales, Yellow B Filter 480 mm.mJ,L (minus blue filter) 1/150-1/250 sec.,j= 21 cm., Gevaert Aerial Film Pan 30. They

PHOTOGRAMMETRIC ENGINEERING726

FIG. 1.

FIG. 2. Cantilever extension vertical errors.

favorable operating conditions, the resultsobtained have usually been discouragingbecause of remaining residual systematicand random errors. Figures 2 and 3 depictthe cantilever errors encountered.

Several novel approaches for extendingcontrol are suggested which should yield im­proved accuracies and, at the same time, givemore reliable results. These techniques, insome instances, are not dependent upon theuse of complicated electronic and opticalmechanisms, or affected by type of terrainor meteorological conditions prevalent. Thespecific technique or combination of meth­ods that would be adopted for a particularproject would be dependent upon suchfactors as the problems inherent with theterrain to be mapped and equipment avail­able. It should be stressed that these ap-

Page 3: Proposed Aerial Triangulation Techniques

PROPOSED AERIAL TRIANGULATION TECHNIQUES

FIG.4a. (at left) Jn flight; FIG. 4b. (at right) Parallel flight.

727

indicated that optimum results were ob­tained with yellow or white targets locatedon a purple-blue background. The sametarget colors were also found best againstnatural ground cover. The background tar­gets were 170 X 180 em. and the centraltargets 20X20 em. (approximately 8 incheson a side). Aircraft or automatically guideddrones could be used to tow targets of thissize without much difficulty. Attached tothe wing or body of the aircraft would be acamera which records the relative positionsof the targets from a horizontal plane.Figure Sb illustrates an exposed print. Allof the mission cameras would be set to ob­tain simultaneous exposures-each expo­sure being automatically triggered by oneof the photo aircrafts. As it is radar couldbe used to measure the distance betweenthe slave aircraft and the towed target.

Some of the advantages of the system are:1. Control is established coincident with

the photographic mission.2. It offers the possibility of setting up

models independently or in sets to ob­tain a best fit. This would be ideallysuited for use on highway, dam, bridge,meteorological and power plant pro­jects.

3. Since each model offers an independentsolution, there is less likelihood ofpropagating errors that are normallyencountered in triangulated strips.

The adj ustmen t process may be ac·complished during or after the aerialtriangulation.

4. An area can be mapped prior to recei ptof ground-control to obtain a relativeorientation. This makes possible im­mediate plotting of maps. Absoluteorientation can be accomplished uponreceipt of known ground-control atsome later date.

5. It can be used with any type of pho­tography, i.e., short or long focal­lengths, vertical, convergent, or ter­restrial views.

Some of the apparent disadvantages ofthe proposed system are:

1. Need for addi tional aircraft and gearif simultaneous exposure is required ofthe slave aircraft.

2. Need for placi ng towed targets in cor­rect position prior to exposure. This,however, is not as critical as the meansfor performing this operation, since thethe targets need only be located some­where in the area of stereo coverage.

CAMERA REVERSAL PHOTOGRAPHY

I n this system aerial photography is ex­posed with the intervalometer of camera setfor 80 per cent forward lap photography.

However, instead of the camera beingpositioned in the same direction throughoutthe en tire mission, every other exposure is

1l0..i10n

LevellndicCl+or'

FIG. Sa. (at left) Slave aircraft; FIG. 5b. (at right) Recording camera.

Page 4: Proposed Aerial Triangulation Techniques

728 PHOTOGRAMMETRIC ENGINEERING

NormClI fli~ht - '01. forword L0l'

7 ~F/i ht Z-------+--

FIG. 6. Camera reversal photography.FIG. 7. Aerial triangulation-normal side lap.

taken with the camera rotated 180 degrees.Thus, two sets of 60 per cent photographywill be obtained, one set being taken in thereversed direction (see Figure 6). Rotationof the camera is made by electrical or me­chanical means. A stop will indicate the 180degree camera-rotation position. Approxi­mately one-half a minute will normally beavailable to rotate the camera before an ex­posure is taken.

The photography obtained would be thesame as if two separate flights were flown,one towards and the other away from an un­controlled area. Thus, procurement of al­most identical duplicate coverage, exposedat the same altitude is obtained to providefor a more rigid triangulation solution. Itwould be exceedingly di fficult and perhapsimpossible to secure duplicate or evenparallel strip coverage over very large dis­tances by conventional methods. Differentweather conditions may also be present inthe return flight to introduce excessive ran­dom errors in the final adjusted positions.

The adjustment of control is based uponthe assumption that the geometry of theinitial oriented model is extended through­out all successive models in the same andadjacent strips by means of common tie-inpass-points. The coordinates of the plottedinstrument points may then be adjusted bydetermining the rate of change of the vari-

ous x, y, and z componen t errors. This is ac­complished by analyzing the common pass­point discrepancies and obtaining a mean fitof the adjusted positions. Figure 7 illus­trates an aerial triangulation with presentphoto strips and Figure 8 utilizes the camerareversal technique.

In bridging or extending control oververy large distances, it is desirable to pro­vide for additional photography beyond theproject limits, to aid in the bz curve solutionand, where possible, to utilize any additionalcontrol to strengthen the adjustment. Theproposed system may then be used to con­siderable advantage, since the trend of thetriangulated strip to and from the uncon­trolled area may be more readily estab­lished.

The results obtained for bridging controlover very large distances by conventionalmethods have indicated the presence of ex­cessive random and systematic errors re­maining in the final adjusted instrumentpositions. Use of the proposed system wouldprovide a convenient tool for averaging-outthese errors. Where no control is available,a map may be constructed wherein the datapresented are relatively correct. An ap­proximate scale may be determined frombarometric altimeter recordings. Availabil­ity of control at a later date would providemeans for establishing absolute orientation.

Loop Traverse OpenTr(1"er.s~

III

-t­I

L--lflight 2

------6. flisht L' I

- 1-- - I

6. I,J ----

FIG. 8. Camera reversal aerial triangulation.

Page 5: Proposed Aerial Triangulation Techniques

-------------------------------------------,

PROPOSED AERIAL TRIANGULATION TECHNIQUES 729

4 i

FIG. 9. Strips triangulated with 15% side lap.

The proposed system is not Ii mi ted to use ofvertical photography; convergent photog­raphy may also be employed. The aerialtriangulation solution may be further en­hanced if si multaneous aerial photography,and if APR and HIRAN data were obtainedin conjunction with atmospheric data. Theproposed system is also ideally suited forhigh altitude, long distance aerial triangula­tion by ei ther stereophotogrammetric oranalytical methods.

60 PER CENT SIDE LAP PHOTOGRAPHY

Experience has indicated that the pass­point control established along the center­line of a strip of triangulated aerial photog­raphy is usually the most accurate. Evi­dently this is so because resolution is bestin this area, relief and tilt effects are mini­mized, and the triangulation process em-

ploys center pass-points for establishingscale. I t follows that the common side-lapareas, usually about 15 per cent (see Figure9), are normally the weakest because of suchfactors as poor resolution, greater distortioneffects, detrimental effects or relief and tilt,and model warpage. The results obtainedare usually unreliable even when commonpass-point values in the area of sidelap areinclose agreemen t prior to adj ustmen t be­cause of the minimum side coverage avail­able.

To take advantage of the more favorableproperties presen tin a stereo model, it isproposed that the photography be flown toobtain 60 per cent side-lap coverage (seeFigure 10). Adoption of this photo patternfor aerial triangulation has the followi ngindicated advantages:

1. 1m prove men t of the accuracy re-

Photo .otr;", n be .OadI" _ <o"l'llcation ".,....

FIG. 10. Flight pattern 60% side lap photography.

Page 6: Proposed Aerial Triangulation Techniques

730 PHOTOGRAMMETRIC ENGINEERING

quired to map large uncontrolled areaswould be considerably aided.

2. Cross-flights could be eliminated orminimized considerably.

3. The need for using wing-points in thestereo compilation operations would beeliminated.

4. The random and systematic errorspresently remaining in triangulatedstrips after block adjustment proce­dures have been performed would bereduced.

5. Stronger radial line and slotted tempJetplots would be permitted.

6. Photo mosaics would be superior inquality since extreme edges of photoswould not have to be used.

7. Flight gaps would be minimized, thuseliminating costly red lights.

Although the proposed system has thedisadvantage of requiring that twice asmany strips be flown, it is felt that the ad­vantages far outweigh this aspect. I n thecompilation phase, it would be necessaryto use only every other strip of photographyas presently practiced. The additional timeneeded to triangulate the extra strips prob­ably would not be significant, since onlycenter pass-points need be precisely tri­angulated. Experience has shown that, inachieving orientation of a stereo model, thegreatest amount of time is consumed in theremoval of minute parallaxes and cross-tiltswhich exist in the corners of the model.These time-consuming operations would bemi ni mized by use of the proposed methodsince only the centerline of the strips needbe used for control purposes. Also it wouldnot be necessary to critically remove minuteparallaxes and maintain elevations in thecorners of each model as is presently done.

CONVERGENT AERIAL TRIANGULATION

A technique for triangulating convergentphotography with projection type instru-

ments, such as the Multiplex, is presentedwhich does not require use of a Stereo­pontometer (see Figure 11). Photos 1F, 2R,and 3R are orien ted relative to each other andto ground-control to achieve absolute orien­tation. Pass-point control is established inthe area of common overlap. Photo 2R isthen replaced with 2F, which is oriented inturn to 1F and 3R. I t should be noted herethat, since 2R and 2F were exposed simul­taneously, there is no air base and 2F needbe oriented only in.p, n, and K. The bx, by,and bz components are not disturbed be­cause of common exposure stations. 4R isthen oriented to 2F and 3R, and pass-pointsare dropped in the area of common overlap.Photo 3R is replaced with 3F and the aerialtriangulation procedure carried forward.Because of the use of triplicate coverageand the large B/H ratio employed, it ap­pears that the proposed technique offers asolution which is simpler than, and may besuperior in accuracy to the present stereo­pontometer method of convergent aerialtriangulation. This technique affords anexcellent means for identifying imagery inthe area of triplicate overlap to achieveoptimum accuracy and reliability of thepass-point connection in a triangulationnetwork.

OPTIMIZATION OF AIRBORNE PROFILE

DATA RECORDING

I t is known that the isobaric surface isnormally more stable and definitive in areaswhere the earth's surface is relatively flat.Tests have also indicated that APR sound­ings are weakest in rugged areas. Therefore,it appears logical that advantage be takenof these phenomena and that considerationbe given to flying APR or radio altimetersoundings over relatively flat terrain, orwhere the terrain characteristics do notchange abruptly in shape and texture. Forexample, flights could be designed to hug

Fig. 11. Convergent Aerial Triangulation.

Page 7: Proposed Aerial Triangulation Techniques

PROPOSED AERIAL TRIANGULATION TECHNIQUES 731

Soundin9' OverFIQt Terrain

Soundin,_ Over Valley AreasAt Optimum Flight Altitude

Soundings OverRU9ged Terral"

FIG. 12. Optimumization of radar soundings.

valley lines and flown at optimum radarflying height (see Figure 12). Results to dateindicate best results at approximately 5,000to 10,000 feet. The control established wouldserve as a basis for setti ng up a pri mary net­work from which intermediate pass-pointdata can be established.

One of the most troublesome areas inaerial triangulation is azimuth determina­tion. The Canadian National ResearchCouncil has successfully used Infrared 60°oblique photography in conjunction withAPR to control the transversal bend of atriangulated strip.2 A standard accuracy of±9 m. over 240 kilometers has been achieved.It is assumed that a collinear relationshipexists between photographic and terrai ndatum planes. The horizon is used to com­pensate for cross-tilt and a straight line isused to align points common to the obliquephotography. Adherence to a straight flightis highly desirable in order to maintain anaccurate systematic adjustment. Because ofthe potential of the Canadian system, theArmy Map Service plans to exploit thistechnique in a forthcoming APR test.

FUTURE PROSPECTS

Many other solutions to this problem areahave recently been advanced to excite theimagination of those interested in triangu­lation processes. Such innovations as theintroduction of the super wide-angle camera,precision automatic coordinate readoutsystems, and high-speed computers foranalytical aerial triangulation operations

2 U. V. Helava, Photogrammetria, XII, 1955­56, #4, pp. 230-235.

hold considerable promise. \\Torld 'vVar I andII concepts of relatively immobile type war­fare have been outmoded with the introduc­tion of high-speed aircraft, missiles, andpentomic armies. I t is anticipated thatphotography and supplemental data willfrequently have to be secured from un­manned, unrecoverable vehicles travellingat high altitudes and speeds, thus suggestingthe need for nonjammable feed-back sys­tems. We have all recently read in our news­papers of the successes in missile recovery.This should open up a vast new area for theattention of map makers.

SUMMARY

The techniques proposed herein, if provenfeasible, may serve to alleviate one of themost pressing problems confronting map­producing agencies-that of securing posi­tional data over sparsely controlled regions.In view of anticipated military demands forrapidly producing maps over these areas,accelerated research and development ef­forts should be made to search out and toresolve new and improved map data record­ing and processing systems. The coincidentapproach, whereby positional data can beobtained concurrently with the procurementof aerial photographic coverage, appearsparticularly worthy of additional attention.With such a capability, speed-up of mappreparation phases would be realized. Aerialtriangulation techniques would also besimplified, since they conceivably could con­cern only single-model adjustments, therebyreducing present complex and time-consum­ing transformation procedures.